On 02/22/2012 08:56 PM, Nicolas Bonneel wrote:
> if you have a parametric surface, then you can compute the curvature
> tensor, and use the one diagonal which is as close as possible to the
> principal eigen vector of the tensor, so as to minimize distortion.
thanks
Analytically the formulae are too much complicated (Riemann theta
functions) so I'd approximate the curvature with the mesh;
is there any book/articles/survey ?
are 9 points enough?
and at the edge?
thanks for any advice